---
title: "Resolve Evaluator Revision"
method: POST
path: "/evaluators/revisions/resolve"
tags: ["Evaluators"]
---

# Resolve Evaluator Revision

`POST /evaluators/revisions/resolve`

Resolve embedded references on an evaluator revision's `data`.

Walks embedded references (for example, references to other
revisions or to secrets) up to `max_depth` and `max_embeds`.
The response includes a `resolution_info` block with counts,
depth reached, and errors according to `error_policy`.

## Request body

- EvaluatorRevisionResolveRequest — Body for resolving embedded references on an evaluator revision's `data`.
  - `evaluator_ref` Reference
    - `version` string, nullable
    - `slug` string, nullable
    - `id` string, uuid, nullable
  - `evaluator_variant_ref` Reference
    - `version` string, nullable
    - `slug` string, nullable
    - `id` string, uuid, nullable
  - `evaluator_revision_ref` Reference
    - `version` string, nullable
    - `slug` string, nullable
    - `id` string, uuid, nullable
  - `evaluator_revision` EvaluatorRevisionInput
    - `evaluator_variant_slug` string, nullable
    - `workflow_variant_slug` string, nullable
    - `evaluator_slug` string, nullable
    - `workflow_slug` string, nullable
    - `evaluator_variant_id` string, uuid, nullable
    - `workflow_variant_id` string, uuid, nullable
    - `evaluator_id` string, uuid, nullable
    - `workflow_id` string, uuid, nullable
    - `variant_slug` string, nullable
    - `artifact_slug` string, nullable
    - `variant_id` string, uuid, nullable
    - `artifact_id` string, uuid, nullable
    - `author` string, uuid, nullable
    - `date` string, date-time, nullable
    - `message` string, nullable
    - `flags` EvaluatorRevisionFlags
      - `is_application` boolean
      - `is_evaluator` boolean
      - `is_snippet` boolean
      - `is_managed` boolean
      - `is_custom` boolean
      - `is_llm` boolean
      - `is_hook` boolean
      - `is_code` boolean
      - `is_match` boolean
      - `is_feedback` boolean
      - `is_agent` boolean
      - `is_skill` boolean
      - `is_chat` boolean
      - `has_url` boolean
      - `has_script` boolean
      - `has_handler` boolean
      - `is_static` boolean
    - `tags` object, nullable
    - `meta` object, nullable
    - `name` string, nullable
    - `description` string, nullable
    - `created_at` string, date-time, nullable
    - `updated_at` string, date-time, nullable
    - `deleted_at` string, date-time, nullable
    - `created_by_id` string, uuid, nullable
    - `updated_by_id` string, uuid, nullable
    - `deleted_by_id` string, uuid, nullable
    - `version` string, nullable
    - `slug` string, nullable
    - `id` string, uuid, nullable
    - `data` EvaluatorRevisionDataInput
      - `uri` string, nullable
      - `url` string, nullable
      - `headers` object, nullable
      - `runtime` 'python' | 'typescript' | 'javascript', nullable
      - `script` string, nullable
      - `schemas` JsonSchemasInput
        - `parameters` object, nullable
        - `inputs` object, nullable
        - `outputs` object, nullable
      - `parameters` object, nullable
  - `max_depth` integer, nullable — Maximum recursion depth when following embedded references. Defaults to 10.
  - `max_embeds` integer, nullable — Maximum number of embeds to resolve. Defaults to 100.
  - `error_policy` 'exception' | 'placeholder' | 'keep'

## Response `200`

Successful Response

- EvaluatorRevisionResolveResponse — Envelope for a resolved evaluator revision.
  - `count` integer — 1 when a revision was resolved, 0 otherwise.
  - `evaluator_revision` EvaluatorRevisionOutput
    - `evaluator_variant_slug` string, nullable
    - `workflow_variant_slug` string, nullable
    - `evaluator_slug` string, nullable
    - `workflow_slug` string, nullable
    - `evaluator_variant_id` string, uuid, nullable
    - `workflow_variant_id` string, uuid, nullable
    - `evaluator_id` string, uuid, nullable
    - `workflow_id` string, uuid, nullable
    - `variant_slug` string, nullable
    - `artifact_slug` string, nullable
    - `variant_id` string, uuid, nullable
    - `artifact_id` string, uuid, nullable
    - `author` string, uuid, nullable
    - `date` string, date-time, nullable
    - `message` string, nullable
    - `flags` EvaluatorRevisionFlags
      - `is_application` boolean
      - `is_evaluator` boolean
      - `is_snippet` boolean
      - `is_managed` boolean
      - `is_custom` boolean
      - `is_llm` boolean
      - `is_hook` boolean
      - `is_code` boolean
      - `is_match` boolean
      - `is_feedback` boolean
      - `is_agent` boolean
      - `is_skill` boolean
      - `is_chat` boolean
      - `has_url` boolean
      - `has_script` boolean
      - `has_handler` boolean
      - `is_static` boolean
    - `tags` object, nullable
    - `meta` object, nullable
    - `name` string, nullable
    - `description` string, nullable
    - `created_at` string, date-time, nullable
    - `updated_at` string, date-time, nullable
    - `deleted_at` string, date-time, nullable
    - `created_by_id` string, uuid, nullable
    - `updated_by_id` string, uuid, nullable
    - `deleted_by_id` string, uuid, nullable
    - `version` string, nullable
    - `slug` string, nullable
    - `id` string, uuid, nullable
    - `data` EvaluatorRevisionDataOutput
      - `uri` string, nullable
      - `url` string, nullable
      - `headers` object, nullable
      - `runtime` 'python' | 'typescript' | 'javascript', nullable
      - `script` string, nullable
      - `schemas` JsonSchemasOutput
        - `parameters` object, nullable
        - `inputs` object, nullable
        - `outputs` object, nullable
      - `parameters` object, nullable
  - `resolution_info` ResolutionInfo
    - `references_used` object[], required
    - `depth_reached` integer, required
    - `embeds_resolved` integer, required
    - `errors` string[]
  - `retrieval_info` RetrievalInfo — References actually used to retrieve a revision. For direct retrievals, `references` carries the artifact / variant / revision that was fetched. For environment-backed retrievals, it additionally carries the environment + environment_variant + environment_revision used to look the target up, and `selector` is {the key: path} map inside the environment's references map that selected the target.
    - `references` object
    - `selector` object, nullable

## Other responses

- `422` — Validation Error

---

[API](https://skmtc.net/agenta-ai/apis/agenta-api.md) · [All operations](https://skmtc.net/agenta-ai/apis/agenta-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/agenta-ai/agenta-api/versions/ea6421d829f1/schema)